first commit

README.md

@@ -0,0 +1,161 @@
+---
+language: fa
+datasets:
+- common_voice
+metrics:
+- wer
+- cer
+tags:
+- audio
+- automatic-speech-recognition
+- speech
+- xlsr-fine-tuning-week
+license: apache-2.0
+model-index:
+- name: XLSR Wav2Vec2 Persian by Jonatas Grosman
+  results:
+  - task: 
+      name: Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: Common Voice fa
+      type: common_voice
+      args: fa
+    metrics:
+       - name: Test WER
+         type: wer
+         value: 11.90
+       - name: Test CER
+         type: cer
+         value: 2.94
+---
+
+# Wav2Vec2-Large-XLSR-53-Persian
+
+Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on Persian using the [Common Voice](https://huggingface.co/datasets/common_voice).
+When using this model, make sure that your speech input is sampled at 16kHz.
+
+The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint
+
+## Usage
+
+The model can be used directly (without a language model) as follows:
+
+```python
+import torch
+import librosa
+from datasets import load_dataset
+from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
+
+LANG_ID = "fa"
+MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-persian"
+SAMPLES = 5
+
+test_dataset = load_dataset("common_voice", LANG_ID, split=f"test[:{SAMPLES}]")
+
+processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
+model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
+
+# Preprocessing the datasets.
+# We need to read the audio files as arrays
+def speech_file_to_array_fn(batch):
+    speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
+    batch["speech"] = speech_array
+    batch["sentence"] = batch["sentence"].upper()
+    return batch
+
+test_dataset = test_dataset.map(speech_file_to_array_fn)
+inputs = processor(test_dataset["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
+
+with torch.no_grad():
+    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
+
+predicted_ids = torch.argmax(logits, dim=-1)
+predicted_sentences = processor.batch_decode(predicted_ids)
+
+for i, predicted_sentence in enumerate(predicted_sentences):
+    print("-" * 100)
+    print("Reference:", test_dataset[i]["sentence"])
+    print("Prediction:", predicted_sentence)
+```
+
+| Reference  | Prediction |
+| ------------- | ------------- |
+| _ | _ |
+| _ | _ |
+| _ | _ |
+| _ | _ |
+| _ | _ |
+
+## Evaluation
+
+The model can be evaluated as follows on the Persian test data of Common Voice.
+
+```python
+import torch
+import re
+import librosa
+from datasets import load_dataset, load_metric
+from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
+
+LANG_ID = "fa"
+MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-persian"
+DEVICE = "cuda"
+
+CHARS_TO_IGNORE = [",", "?", "¿", ".", "!", "¡", ";", ":", '""', "%", '"', "�", "ʿ", "·", "჻", "~", "՞", 
+                   "؟", "،", "।", "॥", "«", "»", "„", "“", "”", "「", "」", "‘", "’", "《", "》", "(", ")", "[", "]",
+                   "=", "`", "_", "+", "<", ">", "…", "–", "°", "´", "ʾ", "‹", "›", "©", "®", "—", "→", "。"]
+
+test_dataset = load_dataset("common_voice", LANG_ID, split="test")
+
+wer = load_metric("wer.py") # https://github.com/jonatasgrosman/wav2vec2-sprint/blob/main/wer.py
+cer = load_metric("cer.py") # https://github.com/jonatasgrosman/wav2vec2-sprint/blob/main/cer.py
+
+chars_to_ignore_regex = f"[{re.escape(''.join(CHARS_TO_IGNORE))}]"
+
+processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
+model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
+model.to(DEVICE)
+
+# Preprocessing the datasets.
+# We need to read the audio files as arrays
+def speech_file_to_array_fn(batch):
+    with warnings.catch_warnings():
+        warnings.simplefilter("ignore")
+        speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
+    batch["speech"] = speech_array
+    batch["sentence"] = re.sub(chars_to_ignore_regex, "", batch["sentence"]).upper()
+    return batch
+
+test_dataset = test_dataset.map(speech_file_to_array_fn)
+
+# Preprocessing the datasets.
+# We need to read the audio files as arrays
+def evaluate(batch):
+	inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
+
+	with torch.no_grad():
+		logits = model(inputs.input_values.to(DEVICE), attention_mask=inputs.attention_mask.to(DEVICE)).logits
+
+	pred_ids = torch.argmax(logits, dim=-1)
+	batch["pred_strings"] = processor.batch_decode(pred_ids)
+	return batch
+
+result = test_dataset.map(evaluate, batched=True, batch_size=8)
+
+predictions = [x.upper() for x in result["pred_strings"]]
+references = [x.upper() for x in result["sentence"]]
+
+print(f"WER: {wer.compute(predictions=predictions, references=references, chunk_size=1000) * 100}")
+print(f"CER: {cer.compute(predictions=predictions, references=references, chunk_size=1000) * 100}")
+```
+
+**Test Result**:
+
+My model may report better scores than others because of some specificity of my evaluation script, so I ran the same evaluation script on other models (on 2021-04-22) to make a fairer comparison.
+
+| Model | WER | CER |
+| ------------- | ------------- | ------------- |
+| jonatasgrosman/wav2vec2-large-xlsr-53-persian | **30.12%** | **7.37%** |
+| m3hrdadfi/wav2vec2-large-xlsr-persian-v2 | 33.85% | 8.79% |
+| m3hrdadfi/wav2vec2-large-xlsr-persian | 34.37% | 8.98% |

config.json

@@ -0,0 +1,76 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
+  "activation_dropout": 0.05,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 1,
+  "conv_bias": true,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": true,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.05,
+  "final_dropout": 0.0,
+  "gradient_checkpointing": true,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.05,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.05,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "mask_feature_length": 10,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.05,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "num_attention_heads": 16,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "pad_token_id": 0,
+  "transformers_version": "4.5.0.dev0",
+  "vocab_size": 67
+}

preprocessor_config.json

@@ -0,0 +1,8 @@
+{
+  "do_normalize": true,
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b859c7f562a2cc3c6002c2eb5178b66777406c4fccf53f196ead46a4f6c4796
+size 1262208535

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

vocab.json

@@ -0,0 +1 @@
+{"<pad>": 0, "<s>": 1, "</s>": 2, "<unk>": 3, "|": 4, "٬": 5, "و": 6, "ـ": 7, "ئ": 8, "ل": 9, "ج": 10, "ک": 11, "R": 12, "ِ": 13, "ع": 14, "َ": 15, "م": 16, "ض": 17, "-": 18, "I": 19, "F": 20, "ذ": 21, "ن": 22, "ژ": 23, "A": 24, "ش": 25, "ث": 26, "Y": 27, "د": 28, "ر": 29, "ّ": 30, "أ": 31, "ق": 32, "ب": 33, "ح": 34, "ظ": 35, "پ": 36, "ت": 37, "خ": 38, "غ": 39, "ط": 40, "ك": 41, "ي": 42, "E": 43, "Ā": 44, "؛": 45, "ی": 46, "چ": 47, "ه": 48, "M": 49, "ف": 50, "آ": 51, "ز": 52, "ص": 53, "س": 54, "گ": 55, "N": 56, "ُ": 57, "T": 58, "S": 59, "Š": 60, "ٔ": 61, "B": 62, "ء": 63, "ً": 64, "ا": 65, "ى": 66}